INDEX

Explanations

words related to technological or political discussions

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ricular

-0.67

estial

-0.61

earchers

-0.61

abase

-0.56

lesiastical

-0.56

ayan

-0.56

reditation

-0.53

tsy

-0.53

 unemploy

-0.53

landish

-0.53

POSITIVE LOGITS

 Both

2.36

Both

2.28

 respectively

2.14

 both

2.03

both

1.99

 Neither

1.88

Neither

1.76

 neither

1.71

 Either

1.44

 latter

1.38

Activations Density 0.870%

words related to technological or political discussions

No Comments

No Known Activations